Google’s AlphaGo to take on world Go champ

Go Google!

Google’s Go-playing program AlphaGo will take on world champion Lee Sedol in a series of games in Korea next month.

There’s a $1 million prize for the winner.

AlphaGo has already bested ranking Go masters in Europe, and now has set its sights on a world championship.

The Go-playing program, developed by Google’s London-based artificial intelligence team DeepMind, trounced the best Go-playing AI programs in a series of matches, as reported by Nature last month.

It is yet another example of how machine intelligence is sweeping one field after another that until recently had been dominated by traditional artificial intelligence approaches.

Rather than processing computer instructions sequentially, machine intelligence uses neural networks — clusters of processors — to handle many tasks simultaneously.

AlphaGo uses these clusters of graphics processors as its neural network brains. The processors, GPUs for short, can handle multiple tasks rapidly and cheaply, which is central to mastering a game with more positions than there are atoms in the universe.

Rather than following strict computer commands, AlphaGo learns from playing the game, like a human. Machine learning, and its subset deep learning, give programs minimal instructions. The results can be seen in robots that learn tasks, self-driving cars, and speech and language programs making their way into mobile applications.

Go is a 2,500-year-old game that was mentioned by Confucius.

Players take turns placing back or white stones on a grid, capturing opponents or surrounding spaces to take territory.

The match will be broadcast on YouTube.

Photo:Google logo (Mercury News archives)


Tags: , , , , ,


Share this Post